The fair use fight over AI training just gained a heavyweight ally for OpenAI. In litigation brought by The New York Times, the Trump administration filed a 20-page brief on September 2 supporting the ChatGPT maker’s right to learn from copyrighted material without paying for it.
The dispute turns on one question with big consequences. Is absorbing books, articles, and other published works into a training run a transformative use, or is it theft? The Times argues the latter. The administration’s brief, submitted in the Southern District of New York, comes down on the side of transformation.
Its arguments lean on national interest. The brief cites an executive order signed last year and warns that a cramped reading of fair use would choke off the innovation behind American AI leadership. Misapplying the doctrine, it says, would slow creative and scientific progress and damage economic mobility.
Recent rulings offer OpenAI encouragement. Judge William Alsup’s $1.5B judgment against Anthropic punished the company for training on pirated shadow libraries, not for the act of training itself. Alsup wrote that an LLM absorbs texts the way a writer studies books, to create something new rather than copy.
An amicus brief cannot decide the case, and the administration has no authority over the dispute. Even so, the filing gives OpenAI’s fair use defense a prominent backer as the suit grinds through discovery.
